Nvidia And CrowdStrike Turned An Attacker Into A Product
CrowdStrike has shipped a commercial cybersecurity product built from a named model designed to emulate attackers. The product, called SafeMind, was announced jointly with Nvidia on stage at CrowdStrike's Fal.Con 2026 conference in Las Vegas on September 1, in front of roughly 10,000 security professionals.
Nvidia CEO Jensen Huang and CrowdStrike CEO George Kurtz presented the launch together. Huang called it "This is the beginning of a new age of cybersecurity.", while Kurtz explained the gap SafeMind is meant to close: "The real gap that I saw was that the attackers had frontier AI, and the defenders didn't."
Two Named Models, Built For Opposite Jobs
SafeMind pairs two purpose-built AI models under one system. Red Tempest is the offensive model, trained to emulate AI-enabled adversaries; Blue Solano is the defensive model, built specifically on Nvidia's Nemotron 3 Super and orchestrated by the larger Nemotron 3 Ultra model.
Kurtz described the pairing directly: "SafeMind brings offensive and defensive models together in a system trained on CrowdStrike's unique cyber data." CrowdStrike Chief AI Officer Dr. Bartley Richardson framed the intended effect on defenders: "Defenders can now act at machine speed."
The Numbers CrowdStrike Put Behind The Launch
CrowdStrike attached three comparison figures to the launch, each measured against frontier general-purpose AI models rather than against its own earlier tools. The company says SafeMind reaches a 29% higher detection rate, remediates threats 6x faster end-to-end, and cuts detection and remediation costs by 99%.
| Metric | SafeMind vs. frontier models |
|---|---|
| Detection rate | 29% higher |
| Remediation speed | 6x faster, end-to-end |
| Detection and remediation cost | 99% lower |
CrowdStrike frames the urgency behind those numbers with two more figures from Nvidia's own blog post: AI-enabled attacks reportedly rose 89% over the past year, and the fastest observed eCrime breakout time - the gap between initial access and lateral movement - is now 27 seconds.
One Vendor's Fifteen Years Of Data, And What Customers Cannot See
Blue Solano's judgment comes from fifteen years of CrowdStrike's own incident-response history, sensor telemetry, and threat intelligence, a training corpus that sits entirely inside one vendor and that no customer outside CrowdStrike can inspect.
That is a familiar trade in enterprise security software, but it lands differently once the product is an AI system making detection and remediation calls at machine speed. A customer adopting SafeMind is asked to trust a model's judgment without visibility into the data that shaped it, an arrangement worth naming plainly rather than treating as incidental to the sales pitch.
What Happens If Red Tempest's Output Leaves The Building
Red Tempest is one of the first widely marketed instances of a frontier cybersecurity vendor selling an attacker-emulation model as a named commercial component, not a research curiosity kept inside a lab.
For EU and UK enterprises and regulators already working through AI Act dual-use classification questions, that is worth examining on its own terms: what happens if an offensive capability built to emulate real adversaries is repurposed, exported outside its intended customer base, or has its outputs leak. Nvidia's own blog post, which confirms the Nemotron 3 Super and Ultra architecture behind Blue Solano, does not name Red Tempest at all; CrowdStrike's press release is the source that puts both models on the record together.
Servola Journal
We do this for everyone trying to keep up with what technology is doing to our lives. The people who build it, and the people it happens to. The Servola Journal exists so that what we learn belongs to all of them.
Nobody pays us for this. No ads, no paywall, free to everyone. We just believe that understanding what's happening to all of us shouldn't depend on who can afford to pay for it.
If it gave you something today, tell us to keep going. Follow us, leave a like, or write a positive comment. We read every one, and they are what keeps us going.
Read next: NYC Bars AI Chatbots for Pre-K to 8th Grade | Mistral's Own Chip Suppliers Just Became Its Owners
